calc_index() adds 1 unit of the level granularity to the expiry passed in parameter to ensure that the timer doesn't expire too early. Add a comment to explain that and the resulting layout in the wheel.
Most-of-the-patch-by: Thomas Gleixner <[email protected]> Tested-by: Juri Lelli <[email protected]> Signed-off-by: Frederic Weisbecker <[email protected]> Cc: Peter Zijlstra <[email protected]> Cc: Anna-Maria Behnsen <[email protected]> Cc: Juri Lelli <[email protected]> --- kernel/time/timer.c | 12 +++++++++++- 1 file changed, 11 insertions(+), 1 deletion(-) diff --git a/kernel/time/timer.c b/kernel/time/timer.c index 2af08a169564..af1c08b0b168 100644 --- a/kernel/time/timer.c +++ b/kernel/time/timer.c @@ -156,7 +156,8 @@ EXPORT_SYMBOL(jiffies_64); /* * The time start value for each level to select the bucket at enqueue - * time. + * time. We start from the last possible delta of the previous level + * so that we can later add an extra LVL_GRAN(n) to n (see calc_index()). */ #define LVL_START(n) ((LVL_SIZE - 1) << (((n) - 1) * LVL_CLK_SHIFT)) @@ -490,6 +491,15 @@ static inline void timer_set_idx(struct timer_list *timer, unsigned int idx) static inline unsigned calc_index(unsigned long expires, unsigned lvl, unsigned long *bucket_expiry) { + + /* + * The timer wheel has to guarantee that a timer does not fire + * early. Early expiry can happen due to: + * - Timer is armed at the edge of a tick + * - Truncation of the expiry time in the outer wheel levels + * + * Round up with level granularity to prevent this. + */ expires = (expires + LVL_GRAN(lvl)) >> LVL_SHIFT(lvl); *bucket_expiry = expires << LVL_SHIFT(lvl); return LVL_OFFS(lvl) + (expires & LVL_MASK); -- 2.26.2
